agentleFS
Sign inStart free
NadavShanun-designNadavShanun-design/opengtm/skills/email-deliverability/SKILL.md

email-deliverability

Skill4 starsChanged 29 days agoMITView on GitHub

Use when cold or lifecycle email is landing in spam, before launching outbound at volume, or when setting up a new sending domain. Covers SPF, DKIM, DMARC authentication, separate sending-domain strategy, inbox warmup, daily volume ramp, list hygiene, bounce and complaint control, the Google/Yahoo bulk-sender requirements, and how to monitor inbox placement. For the copy side of spam avoidance see cold-email-writing.

8 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

7 rules. 0 are common, 4 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep ~2-3 inboxes per domain, and spread volume across domains/inboxes.

    line 43
  2. Distinctive:

    Warm up 2-4 weeks: before real sends - automated warmup exchanges + a trickle of genuine mail.

    line 49
  3. Very specific:

    Ramp volume: start low (~10-20/inbox/day), increase gradually over weeks to a steady cap (commonly ~30-50/inbox/day for cold).

    line 51
  4. Distinctive:

    Verify every address: before sending; keep bounce rate under ~2-3%.

    line 57
  5. Very specific:

    Remove role addresses (info@, sales@), catch-alls you can't verify, and anyone who hasn't engaged across a whole sequence.

    line 59
  6. Very specific:

    Run periodic seed/inbox-placement tests across Gmail/Outlook/Yahoo.

    line 66
  7. Distinctive:

    Do keep bounces <~2-3% and complaints <~0.3%.

    line 73

“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Email deliverability

Overview

The best cold email in the world earns nothing from the spam folder. Deliverability is mostly infrastructure and reputation, not words. Fix the plumbing first; then copy (see cold-email-writing) does the rest.

Three things decide placement: authentication (are you who you say you are), reputation (do mailbox providers trust this domain/IP), and engagement (do recipients open, reply, and not complain).

1. Authenticate - the non-negotiable trio

RecordWhat it provesRule
SPFwhich servers may send for your domainone SPF TXT record, all senders listed
DKIMthe message wasn't altered and is signed by youenable DKIM signing on every sending platform
DMARCwhat to do when SPF/DKIM fail + where to reportpublish a DMARC record; start p=none, move to quarantine/reject

As of the Google/Yahoo 2024 bulk-sender requirements, senders above ~5,000 messages/day to Gmail/Yahoo must have SPF, DKIM, and DMARC, one-click unsubscribe, and keep spam complaints under ~0.3%. Treat these as table stakes for any sender now, not just bulk.

2. Protect your main domain - send from a separate one

  • Never run cold outbound from your primary domain (company.com). Burn it and your invoices, contracts, and real mail go to spam too.
  • Buy lookalike sending domains (getcompany.com, company-hq.com), host the mailboxes there, and redirect them to your main site.
  • Keep ~2-3 inboxes per domain, and spread volume across domains/inboxes.

3. Warm up, then ramp slowly

A brand-new domain/inbox has no reputation. Rules of thumb (not hard law):

  • Warm up 2-4 weeks before real sends - automated warmup exchanges + a trickle of genuine mail.
  • Ramp volume: start low (~10-20/inbox/day), increase gradually over weeks to a steady cap (commonly ~30-50/inbox/day for cold). Don't spike.
  • Pace within the day - even spacing beats a burst at 9:00am sharp.

4. List hygiene

  • Verify every address before sending; keep bounce rate under ~2-3%. High bounces are the fastest reputation killer.
  • Remove role addresses (info@, sales@), catch-alls you can't verify, and anyone who hasn't engaged across a whole sequence.
  • Honor unsubscribes/opt-outs instantly. Keep complaints under ~0.3%.

5. Monitor placement

  • Watch Google Postmaster Tools (domain reputation, spam rate, auth results).
  • Run periodic seed/inbox-placement tests across Gmail/Outlook/Yahoo.
  • Track reply rate as a proxy - a sudden drop often means you're in spam, not that the copy got worse.

Do / Don't

  • Do authenticate (SPF+DKIM+DMARC) before the first send.
  • Do keep bounces <~2-3% and complaints <~0.3%.
  • Don't send cold from your primary domain, ever.
  • Don't ramp fast or blast a cold list on day one - reputation is earned slowly.

Common mistakes

  • DMARC never advanced past p=none, or multiple conflicting SPF records.
  • One inbox blasting hundreds/day → throttled or blocklisted.
  • Buying an unverified list and sending immediately (bounce spike → domain dead).
  • Blaming the copy when the real problem is placement; check Postmaster first.

Go live with Chronos (optional)

Chronos handles the operational side of this for you on your connected inbox: automated warmup, per-inbox daily caps, even per-minute pacing, and send-window timing are built into the sending engine - you don't hand-tune cron jobs.

  • geo_create_campaign / geo_launch_campaign - sends ramp and pace automatically.
  • geo_get_campaign_benchmarks - watch reply/bounce health per campaign.
  • geo_get_connection_page - connect the sending inbox (Gmail/Outlook/SMTP/AgentMail).

You still own domain DNS (SPF/DKIM/DMARC) - set those up as above before connecting.

Connect in ~30s → CONNECT-CHRONOS.md. Free, no card.

See also: cold-email-writing, multichannel-sequences, outbound-metrics-and-benchmarks.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free